无人机影像密集匹配方法  

UAV Image Matching About Dense Point Based on SIFT

作　　者：冯飞 冯建辉 赵艳艳 

机构地区：[1]河北省第一测绘院,石家庄050031 [2]河北省制图院,石家庄050031

出　　处：《遥感信息》2016年第5期122-125,共4页Remote Sensing Information

基　　金：国家创新研究群体科学基金(41021061)

摘　　要：针对无人机航摄影像存在的像幅小、影像倾角过大且倾斜方向没有规律等特点研究了一种基于尺度不变特征变换算法(Scale Invariant Feature Transform,SIFT)特征点匹配结果的无人机航摄影像密集匹配方法。在SIFT特征点匹配结果的基础上,根据立体像对的同名影像相似性原理,结合Moravec特征点提取算法、灰度相关匹配算法和最小二乘匹配算法来获取无人机影像的密集匹配结果,再利用影像密集点匹配结果生成影像覆盖区域的DSM进行精度分析。实验结果表明,基于SIFT特征匹配算法的无人机影像密集匹配方法能够获取可靠的密集匹配结果。The UAV photography has some shortcomings,such as smaller,larger tilt angle of images and the direction of which is irregular.For that reason,this study improves a UAV image matching method about dense point based on SIFT.Firstly,based on the results of SIFT feature point matching,obtain dense point matches combined Moravec and least square image matching algorithm based on the principle of similarity of the same images.Then,use the results of dense point matching to generate DSM in the image coverage area to analysis the result accuracy.Experimental results show that the proposed algorithm can obtain reliable dense point matches.

关 键 词：无人机航摄影像 SIFT 相似性原理 密集点匹配 DSM 

分 类 号：P231[天文地球—摄影测量与遥感]